Training of fuzzy logic systems using nearest neighborhood clustering
L.-X. Wang · 2002
The author first constructs an optimal fuzzy logic system which is capable of matching all the input-output pairs in the training set to arbitrary accuracy. Then an adaptive version of the optimal fuzzy logic system is presented, using the nearest neighborhood clustering algorithm. To do this, clusters of the sample data using the nearest neighborhood clustering algorithm are viewed as sample data and the optimal fuzzy logic system is used as an adaptive controller for nonlinear dynamic systems. The simulation results showed that the adaptive fuzzy controller could produce very good tracking control.>